一种基于CSO-LSTM的新能源发电功率预测方法

doi:10.11896/jsjkx.240600053

摘要/Abstract

摘要： 随着新能源发电技术的快速发展与广泛普及,该类技术已经成为电力系统中关键的一环。其中,对新能源发电功率的准确预测对于电力系统的合理规划有着重要的意义。然而,现有的新能源发电功率预测方法仍存在以下挑战:1)基于深度神经网络的预测模型的超参数对模型的预测性能有着重要的影响,而目前大多数算法仍采用人工确定的方法为超参赋值;2)现有的预测模型难以高效地挖掘时序数据中的长期依赖关系,从而影响预测精度。针对上述问题,本文提出了一种基于CSO-LSTM(Competitive Swarm Optimizer-Long Short-Term Memory)的新能源发电功率预测方法,旨在利用一种两阶段的模型综合地提升预测性能。首先,在模型的第一阶段提出了一种基于竞争群优化的LSTM超参数优化算法,利用竞争群优化算法良好的探索能力和全局优化能力,实现预测模型超参数的自适应调整。然后,在模型的第二阶段设计了一种基于组合多门控机制的LSTM模型,该方法结合自注意力门控机制和组合多个门控网络用于挖掘新能源发电时序数据中的长期依赖关系,从而进一步地适应不同时间尺度下的新能源生成模式。最后,在2个真实数据集和1个仿真数据集上与4个先进的预测方法进行了对比实验,实验结果验证了提出的CSO-LSTM模型的有效性和执行效率。

关键词: 竞争群优化, 长短期记忆神经网络, 新能源发电功率预测, 多尺度时序数据挖掘, 参数优化

Abstract: With the rapid development and wide popularization of new energy generation technology,it has become a key part of the power system.Among them,the accurate prediction of new energy generation power is of great significance for the rational planning of power system.However,the existing new energy generation power prediction methods still have the following challenges:1)The hyperparameters of the prediction model based on deep neural network have an important impact on the prediction performance of the model,and most of the current algorithms still use the artificial method to assign the hyperparameters.2)It is difficult for the existing prediction models to efficiently mine the long-term dependencies in the time series data,thus affecting the prediction accuracy.To solve these problems,this paper proposes a CSO-LSTM(competitive swarm optimizer and long short-term memory) based method for the prediction of new energy generation power,which aims to use a two-stage model to comprehensively improve the prediction performance.Firstly,in the first stage of the model,a LSTM hyperparameter optimization algorithm based on competitive group optimization is proposed,which uses the good exploration ability and global optimization ability of competitive group optimization algorithm to realize the adaptive adjustment of the hyperparameters of the prediction model.Then,in the second stage of the model,a LSTM model based on the combined multi-gating mechanism is designed,which combines the self-attention gating mechanism and the combined multi-gating network to mine the long-term dependencies in the new energy generation time series data,so as to further adapt to the new energy generation patterns at different time scales.Finally,the proposed CSO-LSTM is compared with four advanced prediction methods on two real datasets and one simulation dataset,and the experimental results verify the effectiveness and efficiency of the proposed CSO-LSTM model.

Key words: Competitive swarm optimization, Long-short term memory network, Power prediction of new energy generation, Multi-scale time series data mining, Parameter optimization

中图分类号:

TP870

顾慧杰, 方文崇, 周志烽, 朱文, 马光, 李映辰. 一种基于CSO-LSTM的新能源发电功率预测方法[J]. 计算机科学, 2025, 52(6A): 240600053-11. https://doi.org/10.11896/jsjkx.240600053

GU Huijie, FANG Wenchong, ZHOU Zhifeng, ZHU Wen, MA Guang, LI Yingchen. CSO-LSTM Based Power Prediction Method for New Energy Generation[J]. Computer Science, 2025, 52(6A): 240600053-11. https://doi.org/10.11896/jsjkx.240600053

参考文献

[1]JIA J Y,LIU Y H,LI X J,et al.Research on time sequence prediction of shallow buried tunnel surrounding rock deformation based on improved PSO-BP neural network[J].Journal of Water Resources and Architectural Engineering,2021,19(5):19-22,51.
[2]FENG H,YU C.A novel hybrid model for short-term predictionof PV power based on KS-CEEMDAN-SE-LSTM[J].Renewable Energy Focus,2023,47:100497.
[3]LI Y Z,LIU X L,XING F F,et al.Daily peak load forecasting based on Bi-LSTM and feature correlation analysis[J].Power System Technology,2021,45(7):2719-2730.
[4]XIAO Y,ZHENG K H,ZHENG Z H,et al.Short-term multivariate load forecasting based on multiscale skip-depth long short-term memory network[J].Computer Applications,2021,41(1):231-236.
[5]JIN X,LI L W,JI J N,et al.Short-term power load forecasting based on big data and optimized neural networks[J].Journal of Communications,2016,37(Z1):36-42.
[6]TAN M,YUAN S,LI S,et al.Ultra-short-term industrial power demand forecasting using LSTM based hybrid ensemble learning[J].IEEE Transactions on Power Systems,2019,35(4):2937-2948.
[7]CHAI Y J,ZHANG L J,YAN Y L,et al.Prediction of SRAM read timing based on ALO-BP neural network[J].Electronic Design Engineering,2024,32(8):82-86,91.
[8]SHU X,ZHANG L,SUN Y,et al.Host-parasite:Graph LSTM-in-LSTM for group activity recognition[J].IEEE Transactions on Neural Networks and Learning Systems,2020,32(2):663-674.
[9]ZHANG Z,CHANG L,TIAN M H,et al.Research on Optimizing Transformer Satellite Power Consumption Time Series Prediction Based on CATPCA[J].Journal of Beijing Institute of Technology,2023,43(7):744-754.
[10]HU X T,CHENG C.Time Series Prediction Based on a Multi-Dimensional Cross-Scale LSTM Model[J].Computer Engineering and Design,2023,44(2):440-446.
[11]YIN Y C,TANG J,GU W J.Process production quality prediction integrating temporal knowledge graph and CNN-LSTM[J/OL].https://doi.org/10.13196/j.cims.2023.0324.
[12]PATEL N,VASANI N,JADAV N K,et al.F-LSTM:Federated learning-based LSTM framework for cryptocurrency price prediction[J].Electronic Research Archive,2023,31(10):6525-6551.
[13]LEE C.LSTM-CRF models for named entity recognition[J].IEICE Transactions on Information and Systems,2017,100(4):882-887.
[14]ZHAN X,PAN Z S,LI W,et al.Financial Time Series Prediction Method Based on Multi-Scale Features and Attention[J].Computer Engineering and Applications,2022,58(19):107-115.
[15]CHEN L L,FAN Y C,XIAO H P,et al.HPO-LSTM Prediction Model for Time-Series InSAR Subsidence Monitoring Integrated with Variational Mode Decomposition[J].Nonferrous Metals Science and Engineering,2021,126(3):1-17.
[16]CHEN M M,WANG J F,LI X H,et al.Temporal QoS prediction based on multi-source features and multi-task learning in cloud service recommendation[J].Journal of Sichuan University(Natural Science Edition),2024,61(4):042005.
[17]QIN C K,ZHAO W S,JIA H B,et al.Coal mine microseismic temporal prediction method based on mode decomposition and deep learning[J/OL].http://kns.cnki.net/kcms/detail/11.2190.td.20240328.1143.003.html.
[18]WU J Y,MA G,AI Z T,et al.Spatiotemporal fusion deformation prediction model for rockfill dams and its application in safety monitoring[J/OL].https://doi.org/10.13243/j.cnki.slxb.20230510.
[19]WANG Y,SHEN Y X,MAO S W,et al.LASSO & LSTM integrated temporal model for short-term solar intensity forecasting[J].IEEE Internet of Things Journal,2018,6(2):2933-2944.
[20]ZAN X,LIU Y M,LIU C Y.Traffic Flow Prediction Based on Time-Series Dual-Branch Graph Convolutional Network[J].Journal of Information Systems,2024(1):114-131.
[21]DING H,HU M,YIN J.Research on air traffic flow prediction based on spatiotemporal graph networks[J].Aviation Computing Techniques,2024,54(2):6-9,15.
[22]XU Q,ZHANG L,LIANG C,et al.High penetration new energy grid load forecasting based on joint temporal-spatial scenes and improved TCN[J].Guangdong Electric Power,2024,37(1):1-7.
[23]MENG H,ZHANG T,WANG J,et al.Multi-node short-term electricity load forecasting method based on multi-scale spatiotemporal graph convolutional network and Transformer fusion [J/OL].Power System Technology,2024:1-14.[2024-05-13].
[24]RAN Q,ZHANG Y.Short-term electricity load forecastingbased on modal decomposition and GRU-XGBoost[J].Power Grid and Clean Energy,2024,40(4):18-27,34.
[25]WANG B,WEN Z,ZHOU X,et al.Short-term combined electricity load forecasting model based on variational nonlinear empirical mode decomposition and TCN-TPA-LSTM [J/OL].Modern Electric Power,2024:1-10.[2024-05-13].
[26]WANG Y,HE W,LI J,et al.Distributed new energy collection and regulation strategy based on 5G NR-U[J].Electric Power Information and Communication Technology,2024,22(4):86-91.
[27]JIANG Q,CHANG Y,YI C,et al.Multi-time-scale optimization scheduling method for joint system of new energy generation and pumped storage hydropower station based on NSGA-II serial mode search[J].Journal of Solar Energy,2024,45(4):434-441.
[28]ZHOU Q,OU J,PENG J,et al.Design and implementation of multi-chain blockchain dynamic partition method for microgrid power trading[J].Electronic Components,2020,43(4):771-777.
[29]MA Y,SHI J,GAO J,et al.Power system dispatching method based on big data new energy generation prediction[C]//China Environmental Science Society(Ed.),Proceedings of the 2020 Annual Conference of the China Environmental Science Society(Vol.4).Shandong Institute of Metrology;University of Jinan:Shandong,China,2020.
[30]ZHAO Y,WU L,SONG W,et al..Application of data mining methods in new energy generation[J].North China Electric Power Technology,2015(10):47-51.
[31]GUO H,CHEN Q,ZHENG K,et al.Forecast aggregated supply curves in power markets based on LSTM model[J].IEEE Transactions on Power Systems,2021,36(6):5767-5779.
[32]XIAO T,PEI W,YE H,et al.Time-dependent Correlation-based Ultra-short Term Forecasting of Conditional Probabilistic Power Flow in Distribution Networks[J].Gaodianya Jishu/High Voltage Engineering,2018,44(7):2362-2371.
[33]WANG Z,ZHANG S.Time series prediction model based on time-domain convolution and bidirectional GRU neural network[J].Software Guide,2020,19(3):48-52.
[34]CHEN J,DENG Z,LIU L,et al.Load time series prediction method based on similarity weight coefficient improved Bagging-DNN[J].Electrician Techniques,2023(16):26-29,32.
[35]ZHANG Y,TANG W,ZHANG B.Lithium battery life prediction based on multi-scale decomposition LSTM-ARIMA[J].Journal of Hainan Tropical Ocean University,2024,31(2):59-68.
[36]SUI C,ZHANG L.Short-term load forecasting method based on ConvLSTM-LSTM[J].Electronic Design Engineering,2024,32(10):54-58.
[37]LI Y,LIU X,XING F,et al.Day-ahead peak load forecasting based on Bi-LSTM and feature relevance analysis[J].Power System Technology,2021,45(7):2719-2730.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed